WASHINGTON, Oct 28 (Reuters) – Nvidia (NVDA.O), opens new tab and Palantir Tech (PLTR.O), opens new tab on Tuesday announced a deal in which Palantir will tap Nvidia’s chips and software to help its customers speed up decision making in complex fields such as logistics.
Well-known as a defense contractor, Palantir’s shares have risen this year, driven by increased defense spending on its military-grade AI tools. But the Denver-based company is also pursuing corporate customers, where its platform can ingest data from many different corporate systems, such as staffing and inventory software, to give executives a real-time view of business performance.
Under the deal struck Tuesday and announced at a conference in Washington, Nvidia’s software will be available on Palantir’s platforms. The two said they would help companies with logistics problems, such as when shipments of products from Asia to the U.S. are delayed by storms.
Nvidia technology can generate potential new shipping routes around the storms, and its AI agents can then compare multiple options based on product costs and customer demand in different regions to help select the best one.
“Because of the speed that you do these optimizations, you can then run them hourly to re-optimize your supply chain and rethink changes,” Justin Boitano, vice president of enterprise AI at Nvidia, told Reuters in an interview.
The companies did not disclose the financial terms of the deal.
Kevin Kawasaki, global head of business development at Palantir, said that AI systems on their own can struggle to understand vast streams of corporate data. Palantir’s systems organize the data streams to reflect business conditions in a way that both humans and AI systems powered by Nvidia can use.
“It’s got to be tethered in reality,” Kawasaki told Reuters, referring to how AI systems can help in decision making. “It’s not overly predictive, so much as it’s just it’s very good at recognizing the present. And that’s when you can start making better next-second decisions.”
Source: Reuters